Productive processes are the largest consumers in power systems. The energy required by these processes is usually supplied by the power grid with its associated high operative costs. In this work, we propose a methodology to design energy management systems for self-consumption in productive processes with non-conventional local energy resources. Our goal is to maximize the use of the local energy resources to reduce the amount of energy contracted with the service supplier, and consequently to reduce production costs of the process. This methodology includes a robust-optimization-based energy management strategy to include power variability through the generation of a finite number of possible future scenarios of uncertain variables such as power demand and power from non-conventional energy sources. It allows improving the performance of the power supplies as our simulation results show.